The invention is directed to a method and to an arrangement for accomplishing an interruption-free power supply operation of a system unit according to the preamble of claim 1 or, respectively, 5.
Clusters are computer systems with a plurality of station computers, what are referred to as servers, that use what is referred to as a xe2x80x9cshared peripheral cabinetxe2x80x9d in common. A xe2x80x9cshared peripheral cabinetxe2x80x9d is a system unit that is formed by a system memory component. For example, a disk storage system is such a component.
The components of such clusters can be placed at different locations.
The clusters must be protected against power outage. It is known to connects all devices, i.e. all servers and the xe2x80x9cshared peripheral cabinetxe2x80x9d, to a single, very large, interruption-free power supply.
A corresponding example is known from the Internet in the Czech language on the page http://ns.prosek.cz/ups/emas_soft.html, in English on the page http://www.masterguard.de/asi/8/emas_soft.html and in German on the page http://www.masterguard.de/asi/8/dmas_soft.html. According to the result page of AltaVista(trademark) following an Internet search in all languages for the key word xe2x80x9cPowerProtect NETxe2x80x9d, Point 1, the Czech was most recently updated on Sep. 23, 1997.
It is disadvantageous that a location dependency is established since all components are to be connected to the large, interruption-free power supply. Another disadvantage is that, given outage of the one, large interruption-free power supply, the entire cluster immediately becomes non-functioning. In particular, the commonly used memory component is without a supply current, which can lead to irretrievable data losses. Moreover, an appropriately large, interruption-free power supply is not always commercially available, so that expensive customized systems are needed.
In addition to this Prior Art, IBM Technical Disclosure Bulletin, January 1988, pages 430 through 432, discloses a switch unit that enables a monitoring of the power supply in systems wherein, for example, a plurality of computer units collaborate with a memory component used in common. The individual system components thereby respectively have their own power supply and are respectively connected to the switch unit. By actuating different switches, the switch unit makes it possible to initiate the shut-off of all or individual system component power supplies, whether because there is an emergency are because service work is due to be performed at an individual system component. The disconnect of the power supply is undertaken by the respective, appertaining system component itself. As long as only computer units are affected by shut-offs of the power supply, the other computer units that are still respectively present can collaborate with the shared memory component. What is disadvantageous is that, as mentioned above, each system component has its own power supply and is dependent on the functioning thereof.
Generally interruption-free power supplies are known from the publications Funkschau, 25/96, pages 46 through 49 and eee, No. 16, August 1990, pages 40 through 42.
An object of the present invention is to specify a method and an arrangement for accomplishing an interruption-free power supply operation of a system unit, particularly of a memory component in a computer system of the species initially cited that tolerates errors and can be realized without local linking.
This object is achieved for the method with the method steps according to the characterizing part of claim 1. For the arrangement, the object is achieved according to the features of claim 5.
An interruption-free power supply operation of a system unit in a computer system with a plurality of station computers using the system unit in common that tolerates errors and can be realized without local linking can be realized both with the method as well as with the arrangement.
The independence from local linking and the error tolerance are accomplished in that each station computer is connected on site to its own interruption-free power supply or, respectively, is supplied with power by such a power supply, and that the shared system unit, i.e. the memory component, is connected to respectively one of the active, interruption-free power supplies via a voltage-monitoring, automatic switchover means. In this way, first, an interruption-free power supply of a station computer can fail (network outage/malfunction/service call) and the shared system unit is nonetheless supplied with power. Since each station unit has its own interruption-free power supply, these can also be placed at different locations. The locations, moreover, can lie apart from one another.
Another advantage of the decentralized power supply is that the respective interruption-free power suppliesxe2x80x94since they need not be selected that largexe2x80x94can be in a power class that is commercially available. A voltage-monitoring switchover is likewise commercially available and functions without specific interruption-free power supply monitoring software for multi-server applications. The shared system unit is always supplied with the voltage of an active, interruption-free power supply via the voltage-monitoring, automatic switchover that monitors the input voltages.
Advantageous developments of the invention are the subject matter of subclaims.
With respect to the method, the individual, interruption-free power supplies can be supplied with power via a respectively same phase. A local linking is thereby not established. The computer system remains tolerant of errors as long as an interruption-free power supply and not the supply phase fails. The advantage is that the computer system can be operated when only one phase is available.
When a plurality of phases are available, the individual, interruption-free power supplies can be supplied with power from the various phases. This has the additional advantage that an individual phase can also go down without the entire computer system or, respectively, the shared system component being thereby negatively affected.
A similar effect is achieved when the various interruption-free power supplies are in fact operated not with different phases but with a respectively independently protected identical phase. It is in fact possible that the respectively same phase is present at different locations. But that they are separately protected because they derive from a respectively different feed. I.e., when the phase goes down at one location, it can still be present at the other locations.
The phase redundancy enhances the security and availability of the computer system.